To thoroughly investigate the phenomenon of atypical antipsychotic- associated weight gain, a feeding laboratory paradigm was developed that included obtaining plasma levels of the orexigenic peptide ghrelin that is associated with appetite and eating. This study is a randomized, double-blind, parallel group trial comparing the effects of a 2-week exposure to olanzapine, risperidone, or placebo on plasma ghrelin area under the plasma-time curve (AUC) in 28 healthy human subjects. Subjects were randomized to receive olanzapine, risperidone, or placebo and titrated over 4 days to 10 mg/d or 4 mg/d, respectively. The mean dose at end point was 8.6 + 1.8 mg/d for the olanzapine group and 2.8 + 0.8 mg/d for the risperidone group. Weight changes were significantly different between groups at end point (F2,44 = 10.193; P = 0.0001). The olanzapine group demonstrated a significant increase in weight at end point (2.25 + 1.84 kg) compared with placebo (0.13 + 1.05 kg; P = 0.007). Because of the small subject number, the comparisons between olanzapine and risperidone and risperidone and placebo did not reach statistical significance, although olanzapine's mean weight gain was numerically greater than that of risperidone (2.25 + 1.84 kg vs 1.10 + 0.99 kg) and risperidone's mean weight gain was numerically larger than placebo (1.10 + 0.99 kg vs 0.13 + 1.05 kg). The baseline adjusted Bonferroni corrected contrast of end point ghrelin AUC demonstrated a significant difference between groups (F2,24 = 4.40; P = 0.024), and the post hoc analysis revealed a significant decrease in ghrelin AUC for the olanzapine group in comparison with the risperidone group (P = 0.021) but not between risperidone and placebo or olanzapine and placebo. Ghrelin AUC values did not change significantly from baseline to end point in either of the other 2 groups. The difference between groups approached but did not reach significance (F2,23 = 3.299; P = 0.055) when body mass index change was included as a covariate, suggesting that the difference in ghrelin AUC change followed thechange in body weight. Sedation associated with both active drugs (P = 0.006) and "stuffy nose" associated with risperidone (P = 0.020) were the only statistically different adverse reactions when compared with placebo. Thus, a human feeding laboratory paradigm using a brief exposure to atypical antipsychotics functions as a method to investigate pharmacologically induced weight gain and its association with changes in the orexigenic peptide ghrelin. This rejects the hypothesis that ghrelin levels are elevated by the antipsychotic and that this is a potential cause of the weight gain phenomenon.